PCSK9 is not involved in the degradation of LDL receptors and BACE1 in the adult mouse brain.

Proprotein convertase subtilisin/kexin type 9 (PCSK9) is a secreted protein that regulates hepatic low-density lipoprotein receptor (LDLR) levels in humans. PCSK9 has also been shown to regulate the levels of additional membrane-bound proteins in vitro, including the very low-density lipoprotein receptor (VLDLR), apolipoprotein E receptor 2 (ApoER2) and the beta-site amyloid precursor protein (APP)-cleaving enzyme 1 (BACE1), which are all highly expressed in the CNS and have been implicated in Alzheimer's disease. To better understand the role of PCSK9 in regulating these additional target proteins in vivo, their steady-state levels were measured in the brain of wild-type, PCSK9-deficient, and human PCSK9 overexpressing transgenic mice. We found that while PCSK9 directly bound to recombinant LDLR, VLDLR, and apoER2 protein in vitro, changes in PCSK9 expression did not alter the level of these receptors in the mouse brain. In addition, we found no evidence that PCSK9 regulates BACE1 levels or APP processing in the mouse brain. In conclusion, our results suggest that while PCSK9 plays an important role in regulating circulating LDL cholesterol levels by reducing the number of hepatic LDLRs, it does not appear to modulate the levels of LDLR and other membrane-bound proteins in the adult mouse brain.

mouse VLDLR (2258-VL, R and D Systems) used as positive control for endogenous brain receptor detected by Western blotting was generated from a murine myeloma cell line and included Thr25-Ala798 with a C-terminal 10-His tag (predicted molecular mass of 86.4 kDa, expected 130-175 kDa band in SDS-PAGE under reducing conditions). Recombinant mouse apoER2 receptor (3520-AR, R and D Systems) was generated from Chinese Hamster Ovary cell line and included Asp35-Lys818 with a C-terminal 10-His tag (predicted molecular mass of 87.5 kDa, expected 140-150 kDa band in SDS-PAGE under reducing conditions).
Binding constants were obtained from a series of injections of PCSK9 at a concentration range of 1 nM to 1 µM with a fl ow rate of 30 µl/min. Following injections of PCSK9, receptor-coupled sensor chip surfaces were regenerated with a 6 s injection of 10 mM HCl and a 60 s stabilization period. Data were analyzed using BIAevaluation software. Base lines were adjusted to zero for all curves, and injection start times were aligned. The reference sensograms were subtracted from the experimental sensograms to yield curves representing specifi c binding. Steady-state analysis was used to plot equilibrium binding response (R eq ) against analyte concentration to obtain thermodynamic dissociation constants (K D ).

Mutant mouse line establishment
The Pcsk9 knockout mouse line (PCSK9 KO) was established at the Institut Clinique de la Souris (ICS; llkirch, France). The targeting vector was constructed by successive cloning of PCR products and contained 3.9 kb 5 ′ and 4.3 kb 3 ′ homology arms and a Neomycin selection cassette. Two loxP sequences were located directly at the end of exon 1 and 237 bp before exon 4. The linearized construct was electroporated in C57BL/6N mouse embryonic stem (ES) cells. After selection, targeted clones were identifi ed by PCR using external primers and further confi rmed by Southern blot with Neo internal probe and with 5 ′ and 3 ′ external probes (data not shown). A positive ES clone was injected into BalbC/J blastocysts, and male chimaeras derived gave germline transmission. The resulting line was verifi ed by PCR using external primers, further confi rmed by Southern blot with Neo internal probe (data not shown), and bred with a CMV-Cre deleter model to result in deletion of intron 1-exon2-intron 2-exon 3 and fi rst 1979 bp of intron 3 DNA region. Progeny mice were on C57BL/6 background. RT-PCR was performed with the forward primer (5 ′ CACCATCACCGACTTCAACA 3 ′ ) located on exon 4 and a reverse primer located on exon 5 (GTCACACTT-GCTCGCCTGT). The awaited 72 bp was obtained in the wildtype (WT) and Pcsk9 ϩ/ Ϫ liver samples. In the PCSK9 KO liver samples, no product was obtained, suggesting absence of a knockout RNA containing exon 4-5 (data not shown). Transgenic mice carrying the human PCSK9 gene (PCSK9 TG) were obtained from Jay Horton (University of Texas Southwestern Medical Center, Dallas, TX) and were backcrossed onto a C57Bl/6 background ( 9 ). Brain and liver tissue samples were obtained following transcardial perfusion with ice-cold PBS containing heparin (3 U/ml) and snap-frozen for further analysis. All animals were handled according to the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals guidelines, and culture systems ( 15,16 ). Disruption of the PCSK9-LDLR interaction in cultured human hepatocytes with LDLR subfragments or anti-PCSK9 antibodies results in increased cell surface LDLR levels and cellular LDL uptake (15)(16)(17)(18). Consistent with the role of PCSK9 in the degradation of LDLR in vitro, elevation of circulating PCSK9 levels in mice either through parabiosis or intravenous injection reduces the level of hepatic LDLR ( 9,19,20 ). Injection of PCSK9 into mice also lowers LDLR levels in several extrahepatic tissues, such as the lung, adipose, and kidney ( 19,20 ).
In addition to its role in regulating LDLR levels, there have been recent reports suggesting that PCSK9 regulates cellular levels of two other LDLR family members, the very low density lipoprotein receptor (VLDLR) and the apolipoprotein E receptor 2 (ApoER2). Overexpression of either VLDLR or ApoER2 resulted in increased binding of PCSK9 to CHO-A7 cells ( 21 ). Furthermore, cooverexpression of VLDLR or ApoER2 and PCSK9 in cultured cells increased degradation of these receptors. PCSK9 was also shown to directly bind both VLDLR and ApoER2 in cellfree assays ( 16 ). These results suggest that PCSK9 is capable of binding receptors other than LDLR and promoting their degradation in vitro.
Because VLDLR and ApoER2 are highly expressed in the central nervous system (CNS) and have been previously implicated in Alzheimer's disease (AD) pathology (22)(23)(24)(25)(26), we sought to investigate their potential modulation by PCSK9 in vivo. It has also been recently shown that overexpression of PCSK9 in cells decreased cellular levels of the ␤ -site amyloid precursor protein (APP)-cleaving enzyme 1 (BACE1), a membrane protease responsible for the production of toxic ␤ -amyloid peptides (A ␤ ) that accumulate in neuritic plaques of AD brains ( 27,28 ). In the present study, we investigated whether PCSK9 participates in the regulation of the steady-state levels of LDLR, VLDLR, and ApoER2, as well as BACE1 in the adult mouse brain. We confi rmed that PCSK9 directly binds to LDLR, VLDLR, and ApoER2 in vitro. However, steady-state levels of LDLR, VLDLR, and ApoER2 were similar in the brain of wildtype, Pcsk9-knockout and human PCSK9-overexpressing mice. In contrast with previously published data ( 28 ), we found that the lack or overexpression of PCSK9 in the mouse CNS did not signifi cantly alter BACE1 levels or APP processing to A ␤ . Together, these results demonstrate that while PCSK9 is capable of binding several membranebound proteins in vitro, it does not appear to play a significant role in regulating their steady-state levels in the adult mouse brain.

Recombinant proteins
Wild-type PCSK9 protein was purifi ed from HEK293 generated media as described ( 29,30 ). Recombinant human LDLR, mouse VLDLR, and human ApoER2 used on surface plasmon resonance (SPR) experiments were purchased from R and D Systems. Recombinant carbonic anhydrase was purchased from Sigma, and streptavidin was from Invitrogen. Recombinant neoepitope antibody G2-10 conjugated with alkaline phosphatase. Synthetic A ␤ 40 peptide (American Peptide Co., Inc.) was used as standard and prepared in APP KO brain matrix. After overnight incubation at 4°C, plates were washed and developed using alkaline phosphatase substrate (Applied Biosystems). Luminescence counts were measured using the LJL Analyst (Molecular Devices). Standards run in duplicate were fi t using a third-order spline fi t, coeffi cients were determined, and unknowns were converted to concentrations.

Statistical analysis
Statistical analysis for BACE1 and A ␤ levels between WT and PCSK9 KO mice was performed by Student's t -test.

Direct interaction of PCSK9 with VLDLR and ApoER2 in vitro
Previous studies assessing the interaction of PCSK9, VLDLR, and ApoER2 have generated inconsistent results ( 13,16,21 ). In COS-M cells, PCSK9 binding was only observed when either the LDLR EGF-A domain or a D354L mutation was introduced within the overexpressed VLDLR ( 13 ). However, other studies demonstrated an interaction between PCSK9 and both VLDLR and ApoER2 in CHO-A7 cells ( 21 ). Since direct interaction between purifi ed PCSK9, VLDLR, and ApoER2 proteins was recently demonstrated using a cell-free binding assay ( 16 ), we sought to investigate whether purifi ed PCSK9 bound VLDLR and ApoER2 using SPR ( Fig. 1 ). In agreement with previous reports ( 29,31 ), the K D for PCSK9 binding to LDLR at pH 7.4 was 386 nM ( Fig. 1A ). PCSK9 also bound VLDLR with a similar affi nity (K D = 379 nM) ( Fig. 1B ). In addition, PCSK9 was found to directly bind ApoER2 with a relatively weaker affi nity (K D = 516 nM) ( Fig. 1C ). PCSK9 did not bind either carbonic anhydrase or streptavidin (data not shown), suggesting that the PCSK9 interaction with LDLR, VLDLR, and ApoER2 was specifi c. Together, these data demonstrate that purifi ed PCSK9 can bind to LDLR, VLDLR, and ApoER2 with comparable affi nities in vitro ( Fig. 1D ).

PCSK9 does not regulate LDLR, VLDLR, or ApoER2 levels in the adult mouse brain
Next, we investigated whether PCSK9 played a role in regulating LDLR, VLDLR, and ApoER2 in the CNS, where these receptors are highly expressed. Detergent extracts obtained from the hippocampus and cortex of four-to sixmonth-old male WT, PCSK9 KO, and PCSK9 TG mice were analyzed for steady-state receptor expression levels by Western blotting. While overexpression of human PCSK9 in brain tissue was confi rmed by Western blotting with an antibody specifi c to human PCSK9 ( Fig. 2A ), many commercially available antibodies generated against mouse PCSK9 and additional antibodies raised against human PCSK9 were tested in WT and KO mouse tissue without success in detecting endogenous mouse PCSK9 (data not shown). Due to the unavailability of antibodies that could effectively detect mouse PCSK9 by Western blotting, we the study protocol was approved by the Institutional Animal Care and Use Committee (IACUC).

Taqman analysis
Frozen WT and PCSK9 KO liver samples were homogenized in Trizol (Invitrogen, Carlsdbad, CA), and total RNA was extracted according to the manufacturer's instructions. RNA (100 g) was DNase treated using RNeasy spin columns and RNase-Free DNase set according to the manufacturer's instructions (Qiagen). Total RNA was extracted from frozen WT and PCSK9 KO cerebellum samples using the SV Total RNA Isolation Kit according to the manufacturer's instructions (Promega). The quality and quantity of the isolated RNA was determined with a NanoDrop ND 1000 (NanoDrop Technologies). Total RNA (0.9 g) was reverse transcribed into cDNA in a total volume of 50 µl using Taqman® reverse transcription reagents according to the manufacturer's instructions (Applied Biosystems). A total of 2 µl, corresponding to approximately 36 ng of input RNA, was used in subsequent Taqman® analysis. The ready-made primer probe sets for PCSK9 (Mm01263610_m1) and GAPDH (Mm99999915_g1) were from Applied Biosystems. Reactions were run and analyzed on the Applied Biosystems 7900 thermocycler under the following conditions: 50°C for 2 min, 95°C for 10 min, and then 40 two-step cycles of 95°C for 15 s and 60°C for 60 s.

A ␤ measurement
Mouse brain samples including the cortex and hippocampus were homogenized with 10× volume to weight of 0.2% diethylamine (DEA) buffer (0.2% DEA, 50 mM NaCl) containing Complete protease inhibitor cocktail (Roche). After homogenization, 1 ml of each sample was boiled for 10 min and spun at 14,000 rpm for 1 h at 4°C. The supernatant was neutralized to pH 7.4 with 10% volume of 0.5M Tris-HCL pH 6.8. Black polystyrene plates (Corning Inc.) were coated overnight with 2.5 µg/ml of rodent N -terminal polyclonal antibody (Sig-39153, Covance) as capture antibody, washed, and then blocked with 3% BSA in phosphate-buffered saline. Brain homogenates (100 µl) were added per well in duplicate, followed by 50 µl of A ␤ 40 C -terminal PCSK9 does not affect BACE1 levels or APP processing in the adult mouse brain BACE1 and A ␤ levels were previously reported to be signifi cantly increased in the cortex of PCSK9 KO mice ( 28 ). However, in contrast to these earlier fi ndings, we found no evidence that overexpression or deletion of PCSK9 affected the steady-state levels of BACE1 in the mouse cortex ( Fig. 4A ). We also examined the steady-state levels of fulllength APP and the APP C -terminal fragment ( ␤ -CTF) produced following ␤ -cleavage by BACE1 and before ␥ -cleavage to release A ␤ using BACE1 KO mouse brain as control. In support of our fi nding for the BACE1 levels, APP and APP ␤ -CTF were not altered in PCSK9 KO and PCSK9 TG mice compared with WT controls ( Fig. 4A ).
To further confi rm the lack of effect of PCSK9 deletion on BACE1 levels in the mouse brain, we evaluated a larger set of three-to four-month-old male PCSK9 KO and WT C57BL6 mice (n = 10 per group). By Western blotting analysis, BACE1 levels (detected with two different BACE1 antibodies) and LDLR expression levels were again similar in PCSK9 KO and WT mouse cortex ( Fig. 4B, C ). In the same set of animals, we also found that PCSK9 deletion did not alter the levels of DEA-soluble brain A ␤ 40, the predominant amyloidogenic fragment released after BACE1 and subsequent ␥ -secretase cleavage of APP ( Fig. 4D ). Altogether, the present data suggest that PCSK9 does not compared tissue expression levels of PCSK9 in WT and KO mice by determining PCSK9 mRNA levels using RT-PCR. We observed approximately 87% and 100% reduction in PCSK9 mRNA in KO mice compared with WT for the cerebellum and liver, respectively (data not shown).
The deletion of PCSK9 in KO mice was also confi rmed by examining LDLR levels in the liver. As expected, while PCSK9 TG mice had markedly reduced hepatic LDLR levels, PCSK9 KO mice showed elevated LDLR protein levels in the liver compared with WT mice ( Fig. 2B ). However, in contrast to the liver, LDLR steady-state levels in both the hippocampus and cortex were not altered by either deletion or overexpression of PCSK9 ( Fig. 2A ).
Similar fi ndings were obtained for the other LDLR receptor family members examined ( Fig. 3A , B ). When corrected to the ␤ -actin loading control, steady-state levels of VLDLR and ApoER2 were not altered by PCSK9 deletion or overexpression in the hippocampus and cortex. Recombinant mouse VLDLR and ApoER2 proteins were used to demonstrate the specifi city of the antibodies and confi rm the immunoreactive bands. In summary, these results suggested that despite indication of the direct interaction between PCSK9 and LDLR receptor family members in vitro, altering PCSK9 levels in vivo did not change the steady-state level of LDLR family members in the adult mouse brain. Binding of PCSK9 to LDLR, VLDLR. and ApoER2. A: PCSK9-LDLR binding measured by surface plasmon resonance (Biacore). Representative sensorgram of response units (RU) versus time upon injection of increasing amounts of purifi ed wild-type human PCSK9 over immobilized LDLR ectodomain. B: PCSK9-VLDLR binding measured by Biacore. C: PCSK9-ApoER2 binding measured by Biacore. D: Replot of RU at the end of the association phase versus concentration for purifi ed PCSK9 injected over LDLR (red circle), VLDLR (blue square), and ApoER2 (green diamond). All data were fi t to determine the dissociation constant of the interaction. ApoER2, apolipoprotein E receptor 2; LDLR, low density lipoprotein receptor; PCSK9, proprotein convertase subtilisin/kexin type 9; VLDLR, very low density lipoprotein receptor. lesterolemia and decreased incidence of cardiovascular disease ( 32 ).
In addition to the well-established role of LDLR in regulating circulating LDL levels and, therefore, the risk of CHD, there is increasing evidence suggesting that LDLR and other LDLR family members may also participate in AD pathology, primarily by regulating the levels of apolipoprotein E (apoE) ( 33,34 ). ApoE is the main lipid transporter in the CNS, and the 4 allele of apoE is a wellestablished genetic risk factor for AD ( 35 ). ApoE is known to bind A ␤ , and it colocalizes with amyloid plaques in the AD brain ( 36 ). Recent fi ndings from LDLR transgenic mice suggest that apoE mediates the clearance of A ␤ via binding to CNS LDLR ( 37 ). It has been also proposed that altered signaling by apoE receptors at synapses may underlie the AD risk factor of the apoE4 isoform ( 26,38 ).
appear to play a role in modulating the steady-state levels of BACE1 or the amyloidogenic process of APP in the adult mouse brain.

DISCUSSION
Many studies suggest that inhibition of PCSK9 binding to LDLR and subsequent lowering of circulating LDLcholesterol can be an effective approach to treat hypercholesterolemia. Gain-of-function mutations in PCSK9 are associated with hypercholesterolemia and autosomal dominant hypercholesterolemia (ADH), whereas loss-offunction mutations in PCSK9 are associated with hypocho- Fig. 2. LDLR protein levels were not regulated by PCSK9 expression in the adult mouse brain. The levels of PCSK9 and LDLR in liver and brain homogenates from WT, PCSK9 KO, and TG mice were evaluated by Western blot analysis as described in "Materials and Methods." Equal amounts of sample protein were loaded per lane and actin was used as an additional loading control for all immunoblots. A: Immunoblots showing PCSK9 and LDLR protein levels in the hippocampus and cortex of WT, PCSK9 KO, and PCSK9 TG mice. B: Immunoblots showing PCSK9 and LDLR protein levels in the liver of WT, PCSK9 KO, and PCSK9 TG mice. KO, knockout; LDLR, low density lipoprotein receptor; PCSK9, proprotein convertase subtilisin/kexin type 9; PCSK9 TG, transgenic mice carrying the human PCSK9 gene; VLDLR, very low density lipoprotein receptor; WT, wild type. were no differences in the steady-state levels of these receptors in the brain of PCSK9 KO or PCSK9 TG mice when compared with WT mice, suggesting that PCSK9 is not a physiological regulator of these receptors in the CNS. One possible explanation for these differences is that PCSK9 function may be cell-and/or tissue-specifi c. While PCSK9mediated degradation of endogenous LDLR was observed in HEK293, NIH 3T3, CHO-A7, CHO-K1, Neuro2A, HuH7, and HepG2 cells, no changes in LDLR degradation were detected in COS-1 cells incubated overnight with exogenous PCSK9 or in human fi broblasts overexpressing PCSK9 ( 21,29,41 ). In addition, injection of purifi ed PCSK9 protein or adenovirus-mediated overexpression of PCSK9 in mice resulted in lowering of LDLR protein levels in the liver, lung, kidney, and small intestine, but it had no effect on LDLR levels in the adrenal gland ( 19,20 ). Another possible explanation for our fi ndings is that endogenous PCSK9 expression levels in the brain may not be suffi cient to alter receptor levels. Indeed, PCSK9 mRNA level in the adult mouse brain is much lower than in the liver (data not shown). An alternative pathway for LDLR degradation was recently reported to be mediated by liver X receptors LDLR, LDLR-related protein 1 (LRP1), VLDLR, and ApoER2 are all apoE receptors signifi cantly expressed in the CNS. While LDLR and LRP1 are classic endocytic receptors involved in lipid transport, VLDLR and ApoER2 have been shown to function mainly as coreceptors for the reelin-mediated signaling important for proper neuronal migration to the cortical layers during development ( 38 ). Reelin signaling via VLDLR and ApoER2 also plays a role in the adult brain, where it participates in the regulation of N-methyl-D-aspartate (NMDA) receptor function during synapse formation and maintenance ( 39 ). Since reelin levels were found to be decreased in the brain of AD patients and in a mouse model of amyloidosis ( 40 ), PCSK9 inhibition leading to increased levels of VLDLR and ApoER2 could have the potential to increase reelin signaling and synaptic function in AD patients.
There has been inconsistent data regarding the interaction between PSCK9 and LDLR family members such as VLDLR and ApoER2 ( 13,16,21 ). We have confi rmed by SPR that PCSK9 not only directly binds to LDLR, but also to VLDLR and ApoER2. However, while PCSK9 interacted with LDLR, VLDLR, and ApoER2 in a cell-free assay, there Fig. 4. BACE1, APP, and APP proteolytic fragments were similar in the brains of WT, PCSK9 TG, and PCSK9 KO mice. Brain homogenates from WT, PCSK9 TG, and PCSK9 KO mice were prepared as described in "Materials and Methods." Brain homogenate from BACE1 KO mice was used as control; equal amounts of sample protein were loaded per lane and actin was used as an additional loading control. A: Immunoblots showing full-length APP, APP-CTF fragments, and BACE1 protein levels in the cortex of WT, PCSK9 TG, and PCSK9 KO mice. B: Immunoblots showing BACE1 protein levels determined for additional WT and PCSK9 KO mice using two BACE1 antibodies, ab2077 (Abcam) and EE-17 (Sigma); C: Quantifi cation of BACE1 levels was performed as described in "Materials and Methods" (n = 10 per group). D: Brain (cortex plus hippocampus) A ␤ 40 levels in WT and PCSK9 KO mice were determined by specifi c sandwich ELISA as described in "Materials and Methods" (n = 10 per group). BACE1, ␤ -site amyloid precursor protein (APP)-cleaving enzyme 1; KO, knockout; LDLR, low density lipoprotein receptor; NS, no statistical significance by Student's t -test; PCSK9, proprotein convertase subtilisin/kexin type 9; PCSK9 TG, transgenic mice carrying the human PCSK9 gene; VLDLR, very low density lipoprotein receptor; WT, wild type.
(LXR) agonists via Idol, an ubiquitin-ligase that targets LDLR for degradation by the proteosome ( 42 ). Interestingly, this LXR-induced mechanism did not alter LDLR levels in the liver, suggesting that tissue-specifi c mechanisms appear to contribute to the steady-state levels of LDLR and perhaps other LDLR family members.
A recent study suggested that PCSK9 can modulate BACE1 protein levels in the CNS ( 28 ). BACE1 is the major ␤ -secretase that cleaves APP to generate a soluble fragment (sAPP ␤ ) and a membrane stub ( ␤ -CTF). Subsequent intramembrane cleavage of the ␤ -CTF by the ␥ -secretase complex releases the amyloidogenic A ␤ 40 and A ␤ 42 peptides, which are believed to play a major role in AD pathology. Jonas et al. ( 28 ) reported that overexpression of PCSK9 in cell culture decreased BACE1 protein levels, and conversely, siRNA downregulation of PCSK9 increased BACE1 protein levels in vitro. The authors also showed increased BACE1 steady-state protein levels and increased brain A ␤ levels with PCSK9 deletion in mice. These fi ndings suggested that PCSK9 inhibition as therapy to reduce CHD could lead to increased risk for AD. However, in contrast to the previously published data, we found no changes in the steady-state levels of BACE1 protein in the brain of PCSK9 KO mice compared with WT controls. Furthermore, there were no changes in the overall levels of APP ␤ -CTF or A ␤ 40 peptide levels, which would be expected to be elevated with increased BACE1 activity.

CONCLUSION
Our fi ndings suggest that PCSK9 does not have a role in regulating LDLR family members or BACE1 protein levels in the adult mouse brain and that the development of PCSK9 therapies for CHD is likely not to be hampered by potential CNS adverse effects.